Re: [gentoo-user] "Headless question": Harvesting the results...software needed.

Stroller <stroller@××××××××××××××××××.uk> [14-10-12 09:28]:
>
> On Sat, 11 October 2014, at 3:10 pm, meino.cramer@×××.de wrote:
> > ...
> > Two things need still to be investigated: How can I store GPSdata onto the
> > flash in a way, that no additional data is stored if no movement is
> > there.
>
> This should be pretty easy, but if you want help with it, you're going to need to give us some insight into how you're storing or accessing the GPS data at present.
>
> Having said that, I'm not convinced that this is necessary. I have a logfile here with 91,000 lines of text in it, and it's 9.9mb and bzips down to 369kb. 91,000 is close to 10 entries every minute for a week, and I'd guess that you'd find CSV formatted GPS data written as text to have about the same length and compressibility as a typical /var/log/messages entry.
>
> > And I have to prevent, that a sector/block/organisation unit
> > of the flash is written more than once to prevent ware out of the
> > flash.
>
> Wear-levelling should be handled by the flash memory controller, and you shouldn't need to worry about it.
>
> Stroller.
>
>
>

Hi Stroller,

thank you for your input :) !!

I am still a pure newbie to gps. This morning I discovered, that there
a LOT of kinds of output of the different GPS devices...the output of
mine looks /so standard/, that I thought it would be /THAT STANDARD/.
;)

Ok ... the device I use is an Ultimate GPS by Adafruit Industries.
The device is a GlobalTop-FGPMMOPA6C module with a MediaTek MT3339
chipset.

The output are NMEA sentences (pure ASCII as it seems). The module is
wired to UART01 on my embedded computer, so gpsd can read the data
from /dev/ttyO1.
Via gpsd it is possible to create json output from that (instead of 
NMEA).

The module creates from 1 to 10 position reports per second. I want
to make this speed depended (read: I have to write some code for it,
if I dont find some...). If one choose 1 report/second it will send
5 position report and then 1 status report (current time for example).

Since civil GPS modules have the feature to obfuscate the exact
position up to some degree (otherwise I would build a weapon instead
of tinkering with an embedded system...at least according to some
external laws ;) I see problems in deciding, whether I am really
moving or my position is moving due to """less accuracy of the
chipset""".

Currently the GPS antenna is clamped to my desktop lamp and I am
nearly up to 100% sure that it is not moving at all...but gpsmon gives
me a natural speed of 0.5km/h which is definitely too much for
a warmth induced Brownian molecular motion... ;)

"Wear levelling should be handled by the flash memory controller..."

This "should" makes me a little nervous, Stroller... ;)
Linux includes a filesystem called F2FS, which takes care 
of the wear levelling problem (as far as I know). This F2FS originates
from SAMSUNG, one of the greater manufacturer of Flash memories of all
kind. Why the effort, if the memory controller take care of it?
Stroller, I DONT EXPLICETLY OR IMPLICITELY EXPRESS A "I DONT BELIEVE
YOU" HERE...!
It is just the difference between what manufacturer states ("...should be
handled...") and what mechanisms create more benefits for them.

Best regards,
mcc